Researchers at UCL and the NIHR Moorfields Biomedical Research Centre have developed an artificial intelligence tool, Eye2Gene, that can accurately predict the genetic cause of inherited retinal diseases (IRDs) from routine eye scans.
IRDs are a group of rare monogenic conditions affecting 1 in 3,000 people, with more than 270 different IRD-associated genes identified so far. IRDs cause degeneration of the retina, the light-sensitive tissue at the back of the eye responsible for vision. Some patients with IRDs may be visually impaired from birth, while others experience vision deterioration over time.
A growing number are now being targeted in clinical trials, with approved treatments emerging. However, access to trials and treatments require a genetic diagnosis to be established sufficiently early.
Eye2Gene is a deep learning algorithm developed by UCL researchers to identify IRDs early. Published in Nature Machine Intelligence, the study represents the largest and most comprehensive evaluation of AI for IRDs to date, analysing over 58,000 scans from patients across six international centres (London, Oxford, Liverpool, Tokyo, Bonn and Sao Paulo).
Eye2Gene achieved expert-surpassing accuracy, offering a powerful assistive tool to speed up diagnosis, guide genetic testing, and identify candidates for clinical trials and emerging gene therapies.
Associate Professor Nikolas Pontikos at the UCL Institute of Ophthalmology, Chief Investigator of Eye2Gene, said: “Inherited retinal diseases are a leading cause of blindness in children and young adults but are frequently under diagnosed or misdiagnosed. In spite of advances with genetic testing, accessibility to expertise for administering and interpreting genetic tests is not widespread. We have embedded this expertise into Eye2Gene, an AI developed on the largest expertly curated datasets of retinal scans from genetically confirmed patients with inherited retinal diseases sourced from six international hospitals. This research represents a major advancement in the application of AI to the understudied field of genetic eye diseases and, we hope, a first step towards improving the diagnosis odyssey for patients around the world.”
UCL’s Translational Research Office (TRO) worked closely with Dr Pontikos and his team in advancing this therapeutic innovation towards patient impact.
The model is now accessible online for research use via the Eye2Gene website.